System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

Embed Size (px)

Citation preview

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    1/45

    Optimization based Artificial Neural Networks

    Carlos J. Gmez-Mndez

    University o Puerto Rico at Mayaguez

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    2/45

    Identify a system for which its mathematical model is

    unknown Use Artificial Neural Networks (ANN) to model the system

    rain t e using artic e warm ptimization

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    3/45

    What is s stem identification?

    Mathematical model Example:

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    4/45

    Why identify a system?

    Complicated mathematical model Time consuming to develop the mathematical model

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    5/45

    T es of mathematical models:

    White Box Model

    Black Box Model

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    6/45

    What is an ANN?

    Mathematical model based on biological neural networks Why ANN?

    apta ty

    Non-linear characteristics

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    7/45

    Model of a neuron:

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    8/45

    Structure:

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    9/45

    What is PSO?

    Swarm Intelligence

    Examples: Flocks of birds, School of fish

    Why PSO?

    Iterative

    How it works?

    Particles represent set of data

    Example: X,Y space

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    10/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    11/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    12/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    13/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    14/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    15/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    16/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    17/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    18/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    19/45

    Target is (0,0): x

    10 Particles Random initial positions and velocities

    Global best error and local best errors taken into

    consideration

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    20/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    21/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    22/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    23/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    24/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    25/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    26/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    27/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    28/45

    2

    .

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    29/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    30/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    31/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    32/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    33/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    34/45

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    35/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    36/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    37/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

    .

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    38/45

    2

    1.5

    0.5

    0

    -1

    - .

    -1.5

    -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5-2.5

    -

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    39/45

    Necessary System data is measured

    ANN and PSO parameters are initialized

    Input data is fed to the ANN

    Error is calculated

    ANN is retrained using PSO Repeat until acceptable error is achieved

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    40/45

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    41/45

    The transfer function of the system being modeled with an ANN

    is:

    The in ut to the s stem is an unit ste :

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    42/45

    System response to unit step input and output of initialized

    ANN:

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    43/45

    System response to unit step input and output of trained

    ANN:

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    44/45

    The algorithm developed will be applied for maximum

    power tracking of an ocean wave energy conversion system.

    Preliminary simulations will be done in real-time using

    .

    The algorithms will be usedor power management o a

    DC Motor. The setup consists

    of a DC Motor, DC Generator,

    Power Converter Board, DSP

    Board, and PC.

  • 8/13/2019 System Identification Using Particle Swarm Optimization Based Artificial Neural Networks

    45/45